[PurHur]

Summary

• Validate subject UTF-8 when a preg pattern carries the /u modifier (PCRE2_UTF) before match/split/replace, matching Zend ext/pcre/php_pcre.c.
• Set preg_last_error() to PREG_BAD_UTF8_ERROR (4) and return false/null as appropriate; preg_grep skips bad haystack elements; array preg_replace returns partial results.

php-src reference

• ext/pcre/php_pcre.c — php_pcre_match_impl() UTF-8 validation / PREG_BAD_UTF8_ERROR

PHP implementation

• ext/standard/VmPregPure.php — ensureUtf8SubjectForOpts()
• ext/standard/VmPregPattern.php — PCRE2_UTF constant
• ext/standard/preg_grep.php, ext/standard/VmPreg.php — Zend partial-array behavior

Verification

./script/docker-exec.sh -- bash -lc '
php bin/vm.php test/repro/maintainer_gap_preg_match_invalid_utf8.php
php bin/vm.php test/repro/maintainer_gap_preg_split_invalid_utf8.php
php bin/vm.php test/compliance/cases/pcre/preg_utf8_invalid_subject.phpt
source script/php-env.sh && vendor/bin/phpunit --filter VmPregUtf8SubjectTest
'

Output:

false
4
false
4
match_fail
4
split_fail
4
replace_null
4
array (
  0 => 'a',
)
4
OK (4 tests, 8 assertions)

Closes #17503

Made with Cursor